2 Ways to Correct the Financial Times at AWS (So Far)

AWS has published two blog posts in three weeks titled 'Correcting the Financial Times,' both defending its AI tools after production incidents. The first involved Kiro, AWS's AI coding assistant, executing a CloudFormation teardown in production that took down Cost Explorer in China — Amazon blamed misconfigured access controls and human error. The second involved an engineer following bad advice from an AI tool that had ingested stale internal docs, causing Amazon.com outages. In both cases, Amazon's official line is that AI was not at fault — humans were. The author argues this defensive PR pattern is more revealing than the outages themselves: Amazon is simultaneously laying off engineers, mandating AI adoption, projecting $600B in AWS revenue powered by AI, and then blaming humans when AI-adjacent incidents occur. The real concern isn't AI tool safety — it's a corporate culture that protects AI's reputation at the expense of its engineers.